After analyzing the offensive performances in last season’s Eredivisie in terms of shots, shots on target and goals per shot, now it’s time to do the reverse process and look at the defensive side. Just like in the previous post, data was extracted from the NOS website which presents these parameters in the match reports. The NOS uses data provided by Infostrada Sports, a company that specializes in sports data.
Shots per match
Just like in the previous analysis we’ll start by looking at the amount of shots allowed per match. After all, this is assumed to be one of the key determinants in determining the defensive team performances, and with that, the amount of goals conceded. A future post might dive into exploring the strengths of the relationship between goal conceded and parameters like shots allowed, shot on target percentage and goals per shot on target, but we’ll save that for a later time.
Unsurprisingly, the three teams with the least amount of shots allowed per match are the three top performing teams in the league: PSV (8.9 shots allowed per match), Twente (9.4) and Ajax (9.8). More surprising perhaps is De Graafschap’s sixth place and their 12.1 shots allowed per match, which is about two-third of a shot below the league average. Despite that, only five teams conceded more goals than their 56, which makes it worthwhile to look at the quality of the shots allowed and the shot stopping qualities of the goalkeepers.
Shots on target
The first parameter indicating the quality of the shots allowed would be the fraction of shots that are on target. Take a look at the diagram below and note Twente’s remarkable performance in this department. Only 49.8% of the shots they allowed was on target, whereas the league average was 56.8%. In an average match, given the fact that Twente tends to allow 9.4 shots per match, this would reduce the absolute number of shots on target by 0.65.
At the other end of the spectrum, again, De Graafschap sticks out with almost 61% shots allowed on target, which provides an explanation for their high amount of goals conceded despite their low absolute number of shots allowed.
Goals per shot on target
The third parameter in the chain of conceding would be the amount of goals conceded per shot on target. This reflects a combination of both the quality of the defense and the goalkeeper. These efforts are indistinguishable with the current data, as these rough data do not provide any information on the quality of the shots fired in, other than whether they were on target or not. More elaborate databases, including parameters like shot direction, velocity, shot distance, number of defenders between ball and goal and the position of the goal keeper would allow a form of discrimination between the defenders’ effort and the shot stopping qualities of the goal keeper. Unfortunately, for now, these are not available for the Eredivisie.
Back to the data that we do have then. The single best performing team in terms of goals per shot on target is definitely Ajax. Goal keeper Maarten Stekelenburg (and Vermeer and Verhoeven during some stage of the season) and the Ajax defense allowed only 0.159 goals per shot on target, which was significantly lower than the league average of 0.218, but also significantly lower compared to their rivals PSV (0.195) and Twente (0.214).
Another excellent performance is Groningen’s return of just 0.194 goals per shot on target, which seems an important element to their successful campaign, leading to the club’s best first half of the season ever.
Negative outliers are Willem II (0.306) and VVV (0.269), the two teams finishing 18th and 17th in the final ranking.
Goals per shot
A summary of the data provided in the ‘shots on target’ and ‘goals per shot on target’ diagrams would of course be the aggregated version: ‘goals per shot allowed’. Although this does not provide any additional information to the above, the aggregation of both parameters would still be worthwhile.
Although Twente’s excellent performance in terms of ‘shots on target fraction’ compensates for their average performance regarding the amount of ‘goals allowed per shot on target’, it still doesn’t make up for it. Note that the spread in terms of ‘shots on target fraction’ is much lower than the spread in terms of ‘goals allowed per shot on target’.
Overall, Ajax’ 0.090 goals per shot allowed gave then a competitive advantage over Twente’s 0.107 and PSV’s 0.112. Another thing to note is that Heracles, Feyenoord and De Graafschap all performed above average in terms of the absolute number of shots allowed, but below average in terms of the amount of goals conceded per shot.
The performances in terms of absolute number of shots allowed and amount of goals conceded per shot can be combined into one final figure, displaying both parameters at once. The X-axis represents the amount of goals conceded per shot while the Y-axis represents the number of shots allowed per match. An ideal location would therefore be the bottom left corner of the graph.
The yellow lines indicate the averages for both parameters, making it easy to discriminate whether a team allows too many shots (upper left), allows opponents to score too many goals per shot (lower right), or both (upper right).
In conclusion
A detailed recording of goals, shots and shots on target does provide quite some depth in analyzing the defensive performances of the different teams. And these graphs (at least to me) seem a useful direction to digest this information quickly.
A word of thanks would go out to Infostrada Sports and the NOS, both for collecting and providing the data for this analysis. Further credits are due to a number of people whose work directed me towards the Tableau Software that allow comprehensive data presentations like these. And final thanks go out to a number of excellent ‘soccermetrics’ site that provided inspiration for me and may provide more detailed insight into this subject for those interested : Soccer by the Numbers, A Beautiful Numbers Game, and statDNA’s excellent blog.
bart
28 May 2011
Good overview of defensive parameters with the restricted data. One small thing though …
In the last graph PSV should be LOWER on the Y-axis and further to the RIGHT on the X-axis as their “shots per match” is 8.9 as you stated in the first section (not 11.4 as indicated on the final graph) and the “goals per shot allowed” is 0.112 and not 0.087(or so) as indicated on the graph above. PSV should be closer to FC Twente in the final graph.
I haven’t double checked other teams in the final graph.
Also … a funny fact that these figures could (and I want to emphasize could) point to is that the goalie of FC Groningen (Luciano) is one of the better goalies of the Eredivise. Just recently the 4 men at the table in the RTL TV show Voetbal Inside, Johan Derksen was the instigator of this idea, said what a worthless goalie he is. One of their arguments was his size (height) was not sufficient … (I think this is one of the myths of soccer, that you need a tall goalie, but this point is not important.) Just over half of the shots allowed by FC Groningen were on target, so that’s quite a few, and only a fifth of those went in! Remarkable stats …
Show that gut feeling about a player can be the complete opposite to what stats show … I believe this is important … the opposite can show that many players are actually overrated by teams, fans and press which I believe to be the case quite often in an overhyped sports environment.
cheers,
b.
11tegen11
29 May 2011
That’s very well spotted, thank you. I’ve updated the final figure to position PSV at the correct spot, which is indeed closer to Twente and reflects their status as the team that allowed the least shots per match.
Regarding Luciano, these data do not provide enough depth to fully make the statement that Luciano is one of the best goal keepers of the Eredivisie, but it does prove that Groningen allows a very low number of the opponent’s shots to be on target, and that they conceded a very low number of goals per shot on target.
These data do not provide insight into whether this was due to their quality in defensive positioning (forcing difficult shots), defensive pressure on a shooting opponent, or merely excellent shot stopping. But is indeed highly unlikely that Groningen would achieve these excellently low amount of goals conceded per shot if Luciano was a substandard goal keeper.
I fully agree with you that there is no such thing as potentially misleading as a gut feeling. Once such a feeling exists, the human mind is practically incapable of digesting any information proving that it is false in an appropriate manner. In other words: you see what you think that you see.
Thanks for your comment!
bart
30 May 2011
no problem … and on Luciano …
hehe, here I say that Luciano has some reason to be ranked around the top goalies of the Eredevisie and along comes ADO Den Haag and knocks 5 by him in a complete off day for him and his whole team of FC Groningen (on arguably their most important game of the season) … but I guess the thing learnt from this is that to judge someone through statistics you need a reasonable “sample size” (as they call it in baseball) to be able to take any conclusions from them … but to drop him for the next game is a touch baffling … even though they did win the game (by a spectacular 5-1) but lost out on penalties …